Age estimation using deep learning

Age has always been an important attribute of identity. It also has been an important factor in social interaction. The posture, vocabulary, facial wrinkles and the intonation are all elements that facilitate the prediction of the user's age. Age estimation from the face by numerical analysis finds many potential applications such as the development of intelligent human-machine interfaces and improvement of safety and protection in various sectors such as transport, security and medicine. In many works, researchers are particularly interested in the face's features to regress the age. Recent advances in Artificial Intelligence (AI) and particulary Deep Learning (DL) techniques increase motivations to use this methods to estimate age. In this work, we present a novel method for age estimation from a facial images based on autoencoders. Autoencoder is an artificial neural network used for unsupervised learning of efficient coding. Its aim is to learn a representation for a set of data. The purpose of this work is to exploit the performance of autoencoders to learn features in a supervised manner to estimate user's age. We use MORPH, FG-NET datasets to test the performance of our proposed method. Experimental results show the robustness and effectiveness of the proposed method through the MAE (Men Average Error) rate showing a value of 3.34% for MORPH dataset and 3.75% for FG-NET.